Oligonucleotide−Peptide Complexes: Phase Control byHybridizationJeffrey R. Vieregg,†,# Michael Lueckheide,‡,# Amanda B. Marciel,† Lorraine Leon,§ Alex J. Bologna,†

Josean Reyes Rivera,∥ and Matthew V. Tirrell*,†,⊥

†Institute for Molecular Engineering, University of Chicago, Chicago, Illinois 60637, United States‡Department of Chemistry, University of Chicago, Chicago, Illinois 60637, United States§Department of Materials Science & Engineering, University of Central Florida, Orlando, Florida 32816, United States∥Departamento de Ciencias Biologicas, University of Puerto Rico at Rio Piedras, San Juan, Puerto Rico 00925, United States⊥Institute for Molecular Engineering, Argonne National Laboratory, Argonne, Illinois 60439, United States

*S Supporting Information

ABSTRACT: When oppositely charged polymers are mixed,counterion release drives phase separation; understanding thisprocess is a key unsolved problem in polymer science andbiophysical chemistry, particularly for nucleic acids, polyanionswhose biological functions are intimately related to their highcharge density. In the cell, complexation by basic proteinscondenses DNA into chromatin, and membraneless organellesformed by liquid−liquid phase separation of RNA and proteinsperform vital functions and have been linked to disease.Electrostatic interactions are also the primary method used forassembly of nanoparticles to deliver therapeutic nucleic acidsinto cells. This work describes complexation experiments with oligonucleotides and cationic peptides spanning a wide range ofpolymer lengths, concentrations, and structures, including RNA and methylphosphonate backbones. We find that the phase ofthe complexes is controlled by the hybridization state of the nucleic acid, with double-stranded nucleic acids forming solidprecipitates while single-stranded oligonucleotides form liquid coacervates, apparently due to their lower charge density. Addingsalt “melts” precipitates into coacervates, and oligonucleotides in coacervates remain competent for sequence-specifichybridization and phase change, suggesting the possibility of environmentally responsive complexes and nanoparticles fortherapeutic or sensing applications.

■ INTRODUCTION

In addition to storing information in their sequence, nucleicacids are among the most highly charged polymers known(axial charge density of 6 e−/nm for double-stranded B-formDNA), and interact strongly with other charged molecules inthe cell. A striking example of this is chromatin formation, inwhich genomic DNA is condensed several thousand-fold intoμm-size chromosomes, a process mediated by basic histoneproteins and polyamines.1,2 RNA plays a key role in manymembraneless organelles (also known as “biomolecularcondensates”), intracellular phase-separated droplets assembledat least partially by electrostatic interactions. These areimplicated in numerous essential cellular processes, includingregulation of gene expression, embryonic development, andmetabolism, and have recently been linked to diseases ofaggregation such as Alzheimer’s and ALS.3−6 Phase separationcan also occur when individual nucleotides are mixed withsimple polycations, and the resulting droplets readilyencapsulate oligonucleotides, an observation of great interestto early life scenarios and the “RNA world” hypothesis.7

Understanding the interactions of long, double-stranded DNA

(dsDNA) with cationic small molecules and polymers has beenone of the classic problems of biophysical chemistry for nearly60 years, but, despite its importance, we know far less about thebehavior of short or single-stranded nucleic acids, a deficit thatmotivates the present study.Electrostatic interactions are also the dominant tool used to

assemble therapeutic nucleic acids for delivery into cells in vitroand, increasingly, in vivo. Synthetic polycations such aspolyethylenimine and polylysine (pLys) readily form poly-electrolyte complexes with nucleic acids, and have been usedfor many years to deliver dsDNA to cells, as well as morerecently to encapsulate large DNA nanostructures.8−10 If thepolycation is conjugated to a neutral hydrophilic block such aspolyethylene glycol (PEG), then phase separation occurs onthe nanoscale, producing “polyelectrolyte complex coremicelles” that show great promise for gene delivery invivo.11,12 Oligonucleotides and RNA should also be amenableto polyelectrolyte delivery, but our knowledge of even basic

Received: April 9, 2017Published: January 9, 2018

Article

pubs.acs.org/JACSCite This: J. Am. Chem. Soc. 2018, 140, 1632−1638

© 2018 American Chemical Society 1632 DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

Dow

nloa

ded

via

UN

IV O

F C

HIC

AG

O o

n O

ctob

er 2

2, 2

018

at 1

6:52

:20

(UT

C).

Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

structure−property relationships is incomplete, with differentresearch groups reporting widely divergent results when RNAoligonucleotides are complexed with cationic peptides.8,12−17

Improved knowledge of the complexation properties ofoligonucleotides is therefore required to enable rational designof more effective delivery vehicles.In aqueous solution, long dsDNA forms phase-separated

polyelectrolyte complexes when mixed with cations with charge≥3 or with cationic polymers. The dominant driving force forcomplexation is entropy gain from release of low-valencecounterions;18 at low DNA concentrations, Poisson−Boltz-mann models accurately predict key quantities such as thefraction of DNA charge that must be neutralized by polycationsfor phase separation to occur.19 Upon complexation, dsDNA oflength ∼700 base pairs (bp) or more forms toroidal solids withthe DNA helices wrapped circumferentially, while shorterdsDNA forms disordered rod-like precipitates, presumablybecause of a prohibitive energetic cost to bend the stiff DNA(persistence length 50 nm ≈ 150 bp) into smaller structures.1,20

Complexation has also been shown to stabilize the DNAdouble helix against thermal denaturation and allow formationof normally unfavorable structures such as Z-form and triplexDNA, presumably by reducing electrostatic backbone repul-sion.1,21

The complexation behavior of oligonucleotides (N < 100 nt),single-stranded nucleic acids, and RNA has not been studied toa comparable degree, and may differ substantially from longdsDNA. The electric field around a charged polymer decreasesnear its end, and thermodynamic measurements show thatthese “end effects” can extend for as much as 10 nt,22,23

suggesting that shorter oligonucleotides may behave much lesslike polyelectrolytes than longer ones. Single-stranded nucleicacids are more mechanically flexible and have a lower chargedensity than double-stranded DNA, and RNA duplexes (whichadopt an A-form helical structure, vs B-form dsDNA) havebeen shown to be complexed much less effectively by the smallcations cobalt hexamine and spermine than is DNA.24,25 Nosystematic understanding of these effects on complexation andphase separation currently exists, despite the importance ofthese molecules to both natural and synthetic systems.Polymer science provides some, albeit incomplete, guidance

for understanding polyelectrolyte phase separation. In 1929, deJong reported the separation of gum arabic (a carbohydratepolyanion) and gelatin (a polycation below pH 4.8) solutionsinto polymer-rich and -poor liquid phases upon mixing, aphenomenon they named “complex coacervation”.26 Subse-quent investigations revealed this to be a general occurrencewhen oppositely charged polyelectrolytes of sufficient chargeand length are mixed; as with dsDNA, entropy gain fromcounterion release drives complex formation and phaseseparation, resulting in either hydrated liquids (coacervates)or solid precipitates.27−29 Polyelectrolyte complex materials arewidely used in industry,27,30 despite an incomplete under-standing of the molecular basis of their properties. Severalmodels have been developed to describe the complexationprocess, differing in the relative importance of ion pairing,hydration effects, polymer structure, and electrostatics.28,29,31,32

In many cases, detailed molecular features such as hydrogenbonding and chirality also play key roles in determiningcomplex behavior.33 Despite significant advances in describingcertain systems, basic properties of polyelectrolyte complexes(e.g., the phase: liquid coacervate vs solid precipitate) cannotcurrently be predicted from the structures of their components.

This manuscript describes results of an investigation of theeffects of polymer length, structure, concentration, and saltconcentration on the complexation of oligonucleotides bycationic peptides and polyamines. The use of defined sequenceoligomers enables exploration of a large, yet well-definedparameter space in molecular charge, length, and structure. Byworking at relatively high concentration, we are able to visualizethe complexes directly, enabling a clear determination of theirproperties. These differ in several interesting ways from thoseobtained with longer polymers, and should inform both basicquestions of biopolymer complexation and delivery oftherapeutic nucleic acids.

■ RESULTSFigure 1A shows typical micrographs of complexes formed by22 nt oligonucleotides and 50 amino acid pLys peptides mixedat equal charge (amine and phosphate) concentration. Astriking qualitative difference is immediately apparent: double-stranded oligonucleotides form irregular solid precipitates,while single-stranded oligonucleotides form spherical coac-ervate liquid droplets. Upon mixing, we observe very rapid

Figure 1. Oligonucleotides and poly(L)lysine (pLys) form phase-separated complexes upon mixing. (A) 22 nt single-stranded DNA and50 aa pLys form liquid droplets when mixed at 2.5 mM amine andphosphate concentration. 22 bp double-stranded DNA and 50 aa pLysform solid precipitates when mixed under the same conditions. Imagestaken 4 h after mixing. (B) Quantification of noncomplexed DNAshows that the complexes appear nearly neutral (black line) regardlessof bulk charge ratio and polymer length: [N]/[P] ≡ [pLys amines]/[DNA phosphates]. Total charge ([amine] + [phosphate]) is fixed at 5mM. Solution DNA values are normalized to 1 at [pLys] = 0 and 0 at[DNA] = 0. (C) Phase separation is consistent across a wide range ofpolyanion: polycation concentration ratios ([N]/[P] = 1 shown inPanel A).

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1633

(seconds) formation of small droplets and precipitates, whichthen coalesce into larger aggregates through Browniandiffusion. Droplets containing single-stranded DNA complexesundergo rapid hydrodynamic relaxation upon coalescence(movie Coacervate coalescence.avi of the Supporting Informa-tion, SI), while the dsDNA precipitate complexes stick togetherwithout significant changes in component shape (moviePrecipitate coalescence.avi). The phase of the complexes isinsensitive to the order of polymer addition, although the sizedistribution varies somewhat with the rates of addition andmixing. Fluorescence recovery after photobleaching (FRAP)measurements show recovery for the coacervate droplets butnot for the precipitates (SI Figure S1), consistent with liquidand solid phase formation, respectively.We observed the same phase behavior across all polymer

length combinations and charge ratios measured (Figures 1A,C,S2, and S3), spanning nearly 2 orders of magnitude in bothpolymer length and charge ([amine]/[phosphate], N/P) ratios.Under our experimental conditions, the complexes’ (liquid andsolid) size stops increasing after 30−60 min, presumably due tocoalescence of all phase-separated material in the local area ofeach complex. Quantification of the images (SI Figure S4)shows that droplets appear largest at bulk neutrality (N/P = 1)but that the droplet size is largely independent of polymerlength when mixed under identical conditions. The irregularshape of the dsDNA precipitates makes accurate quantificationdifficult, but the trends appear similar. Phase behavior was alsoidentical at all concentrations we measured, from 0.1 to 10 mMtotal charge (SI Figure S5), although the complex size and rateof growth increased with concentration.Centrifuging the samples collects the mixtures into two

macroscopic phases, as observed for other complex coac-ervates.27 Quantifying the amount of DNA remaining in thesupernatant thus provides a measure of the fraction of DNAincorporated into phase-separated complexes and (when pLysis the limiting reagent) the complex stoichiometry. As shown inFigure 1B, we observe apparently stoichiometric incorporationof nucleic acid into phase-separated complexes regardless of thelength of either polyelectrolyte or the bulk charge ratio. Fouriertransform infrared spectroscopy (FTIR) measurements (SISection 2.2 and Figures S6 and S7) indicate that double-stranded DNA maintains its B-form helical structure in theprecipitate complexes, while strong lysine-phosphate inter-actions are observed for both types of complexes. No additionalabsorbance peaks are noted upon complexation of the single-stranded oligonucleotides, consistent with the idea thathybridization and complexation are largely independent at themolecular length scale.

■ EFFECT OF SALT AND TEMPERATURESalt ions play a key role in polyelectrolyte behavior, as theyscreen electrostatic interactions. Figure 2A shows typical results(see also SI Figure S8) for the single-stranded complexes withincreasing NaCl concentration: the coacervate droplets firstswell, then shrink and ultimately dissolve for salt concentrationsgreater than a critical value. Double-stranded complexes displaymore complex phase behavior: as salt concentration isincreased, the precipitates (Figures 2B and S9) become“softer”, with rounded edges. Continued increase in saltconcentration produces a transition from solid precipitates toliquid coacervates, followed by eventual dissolution as for thesingle-strand complexes. This precipitate−coacervate transitionis unlikely to reflect denaturation of the DNA duplex, as salt

Figure 2. Oligonucleotide−peptide polyelectrolyte complex behaviorvs salt concentration and temperature. (A) Representative phase-contrast images of complexes (2.5 mM each in amines andphosphates) formed between single-stranded DNA and polylysine(scale bar 50 μm). At very low [NaCl], large numbers of micron-sizedcoacervate droplets form. At intermediate concentrations, droplets arelarger, and then shrink again before dissolving (not shown) as [NaCl]increases further. (B) Complexes between double-stranded DNA andpolylysine form precipitates at low and moderate [NaCl]. For longerpolymers, the precipitates appear softer, with rounded edges, as acritical [NaCl] concentration is approached. Above this concentration,liquid droplets are observed; these eventually dissolve at concen-trations similar to those required for single-stranded DNA of the samelength. (C) Phase diagram for oligonucleotide−pLys complexation.Complexes with single-stranded (left bar of pair) and double-strandedDNA (right bar) show similar stabilities, which increase with polymerlength. By contrast, the precipitate/coacervate transition concentration

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1634

stabilizes nucleic acid hybridization,34,35 and has been observedwith synthetic polymers and peptides, albeit at lower saltconcentrations.36,37

Quantifying the salt concentrations required for theprecipitate−coacervate transition and for dissolution over arange of oligonucleotide and pLys lengths produces the phasediagrams shown in Figure 2C. We find that the stability of thecomplexes against dissolution is quite similar for the single-stranded and double-stranded complexes, despite their differ-ences in microscopic structure and macroscopic phase. Forboth types, stability increases dramatically with the length ofeither polymer (from 200 mM NaCl for the 10 nt/10 aacomplexes to 1.2 M for the 88 nt/100 aa complexes). Theprecipitate−coacervate transition, by contrast, occurs within anarrow range of salt concentrations (500−700 mM) thatappears only weakly dependent on polymer length. Measuringthe amount of DNA in the supernatant (Figure 2D) shows thatthe complexes’ shrinking and dissolution at high saltconcentrations reflects loss of DNA to solution, but that theprecipitate−coacervate transition is not accompanied bysignificant nucleic acid release. The images shown in Figure2A−D are for complexes mixed at the stated conditions; weobserve similar behavior, including rapid “melting” ofprecipitates into coacervates, when concentrated salt solutionsare added to preformed complexes (SI Figure S10, movie NaClmelting move.avi).We also investigated the effect of temperature on the

complexes by mixing them at temperatures up to 55 °C and byheating and cooling preformed complexes. The single-strandedcomplexes showed no visible change with temperature,however we observed a precipitate−coacervate transition at∼50 °C for complexes containing the shortest (10 bp) double-stranded DNA at 300 mM NaCl (Figure 2E). The transition isreversible and rapid (within a minute upon temperaturechange). The observed transition temperature is very similarto the melting temperature of the 10 bp dsDNA duplex in theabsence of polycations (SI Table S4), suggesting that thecomplexes may be responding dynamically to disruption ofoligonucleotide base pairing rather than changing phase due toweakened electrostatic interactions. Consistent with thishypothesis, we observed no precipitate−coacervate transitionsfor complexes containing longer oligonucleotides, whosemelting temperatures are higher than 55 °C, the highesttemperature we could access.

■ POSTCOMPLEXATION BEHAVIORWe investigated whether the single-stranded DNA in thecoacervate complexes remained competent for hybridization by

either mixing separately formed complexes or by addingadditional DNA to existing complexes. Results with preformedcomplexes are shown in Figure 3: droplets containing non-

complementary DNA mix readily to form larger spherical liquiddroplets (movie Panel A, Non-complementary fluor movie.avi).By contrast, when droplets containing complementary DNAtouch, they stick together without hydrodynamic relaxation toform irregularly shaped, solid aggregates in which individualdomains can be followed for long times (movie Panel B,Complementary fluor movie.avi). After agitation, theseaggregates appear very similar to those formed by mixingpLys with prehybridized double-stranded oligonucleotides (SIFigure S11). Interestingly, both the liquid and solid complexesdisplay noticeable FRET signal (SI Figure S12), implying closeproximity of oligonucleotides even (for the coacervates) in theabsence of hybridization. Adding free DNA to preformedcoacervate droplets showed a similar result (SI Figure S13): thecomplexes undergo a liquid-to-solid phase transition only uponaddition of complementary oligonucleotides.

■ DETERMINANTS OF COMPLEX PHASETo determine what controls the phase of the complexes, wecarried out several lines of experiments in which one or theother polymer was modified structurally or chemically.Changing the chirality or charge density of the cationic peptidedid not affect the phase of the complexes, and the samebehavior was also observed with the polyamines spermidine(3+) and spermine (4+) (SI Figures S14−S16). Single- anddouble-stranded RNA oligonucleotides displayed the samephase behavior as DNA (Figure 4): single-stranded RNAformed coacervates and double-stranded RNA formedprecipitates.

Figure 2. continued

(split in right bar) is only weakly dependent on polymer length. (D)Fraction of DNA complexed vs [NaCl] and polymer length for single-and double-stranded DNA (66 nt/bp shown for clarity). Oligonucleo-tides are released to solution as observed complex size shrinks, butlittle loss is observed during the precipitate−coacervate transition: seesolid orange (30 aa pLys; 500−600 mM NaCl), solid green (50 aapLys; 500−700 mM NaCl) curves. Values are normalized to theaverage value at 1 M NaCl to aid visual comparison. (E) At 300 mMNaCl, complexes between 10 bp double-stranded DNA and pLysundergo a melting transition at ∼51 °C: solids with rounded edges areobserved just below this temperature and spherical liquids areobserved at higher temperatures. Scale bar = 25 μm.

Figure 3. Coalescence of coacervate droplets formed from non-complementary and complementary DNA oligonucleotides. Coac-ervate complexes are formed separately with TAMRA (red) andfluorescein (FAM, green)-labeled 22 nt oligonucleotides and 50 aapLys. At T = 0, FAM-labeled coacervates are added to microplate wellscontaining the TAMRA-labeled coacervate droplets. (A) Droplets withnon-complementary DNA sequences merge into larger spheres withuniform color and remain liquid. (B) Droplets with complementaryDNA sequences stick on contact, forming patchy, irregular solids inwhich individual domains can be followed for long times. Movies ofboth processes are available in the SI.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1635

We next investigated partially hybridized DNA duplexes,either by annealing complementary oligonucleotides of differ-ent lengths or by designing hairpins with a constant total lengthof 44 nt and double-stranded stems of 5−20 bp. Both systemsgave similar results (SI Figure S17): precipitates were formedwhen the DNA was 40% or more double-stranded andcoacervates formed when lower fractions were hybridized.Finally, we prepared reduced-charge DNA analogs by

substituting uncharged methylphosphonate (MP) linkages forphosphodiesters at 10 positions in a 22 nt oligonucleotide andits complement and preparing complexes with 50 aa pLys.Figure 4B, D shows the results: DNA−DNA and DNA−MP(77% of total DNA charge) duplexes produced precipitates, butthe MP−MP duplex (55% charge) produced coacervates thatwere visually indistinguishable from those formed by single-stranded DNA. UV melting confirmed that duplexes wereformed in both combinations (SI Figure S18), indicating thatthe phase difference is due to the presence of the MPsubstitution rather than lack of hybridization.We also designed a range of oligonucleotide sequences to

determine whether the observed phase behavior is sequence-dependent. While most sequences formed coacervates as singlestrands, a few (3/14, 21%) formed solid complexes withouttheir complement being present. Further investigation (SISection S3) suggests that this is due to either dimerization as aresult of partial self-complementarity or to a high density ofpurine nucleotides, which is known to promote noncanonicalfolded structures, particularly in the presence of polycations.1,21

■ DISCUSSIONOur measurements show that the phase of complexes formedbetween nucleic acids and cationic peptides is controlled bynucleic acid hybridization over nearly 2 orders of magnitude inpolymer length ratio, charge ratio, and total concentration, aswell as with mixed lysine-glycine peptides and small poly-amines. To our knowledge, this is the first systematicinvestigation of this phenomenon, particularly in connectionwith hybridization. As discussed earlier, long double-strandedDNA is known to form solid complexes when mixed withpolycations or polyamines, but we are aware of few reportscharacterizing coacervate formation by nucleic acids of any

length or structure, and our results may help explain severalpuzzling results from the past several decades, as well asinforming understanding of liquid phase separation in cells. In1969, Shapiro et al. described phase separation of long dsDNAand polylysine into highly hydrated, spherical aggregates whenmixed at monovalent salt concentrations of ∼1 M.38 In light ofthe solid−liquid transition with increased NaCl we observe forlonger oligonucleotides, it seems reasonable to suspect thatthese aggregates were coacervate droplets “melted” by the highsalt concentration. In 1979, Porschke reported a turbidsuspension, ascribed to coacervation, when trilysine ortriarginine was mixed with long (N > 750) polyribonucleo-tides,39 but the phase of the complexes was not determinedexperimentally. Aumiller and Keating observed coacervateformation when poly(U) or yeast tRNA was mixed witharginine-rich cationic peptides or polyamines.40,41 Recently, Yinet al. reported microdroplet formation upon mixing ofpolylysine and 21 nt ssDNA.42 Our results suggest thatcoacervation is likely a general phenomenon for single-strandednucleic acids complexed with cationic polymers.Interestingly, the poly(U)−spermine coacervates were only

observed above a critical temperature, likely connected todenaturation of single-stranded stacking interactions in the long(2000−3000 nt) polyribonucleotide. Temperature cyclingproduced reversible phase change, similar to our results withthe 10 bp DNA complexes. Similarly, Jain et al. recentlypublished results showing that partially complementary triplet-repeat RNA sequences associated with Huntington’s diseaseand ALS form coacervate droplets in the presence of Mg2+cations, followed by gelation of the droplets ascribed toformation of intermolecular base pairs.6 These results areconsistent with our observation of solidification whencoacervate droplets containing complementary oligonucleotidesare mixed, and suggest a unified picture in which nucleic acidbase pairing interactions control the dynamics of phase-separated bodies formed through electrostatic interactions.The ability of nucleic acid polyelectrolyte complexes to responddynamically to changing environments and base pairing statesmay be important for their role in living cells, as well asproviding opportunities for engineering responsive moleculardevices and sensors.Single- and double-stranded nucleic acids differ in several

ways that might explain the link between hybridization andphase: double-stranded DNA has a charge density ∼2.4 timeslarger than the single-stranded form, is less flexible (persistencelength 50 nm vs ∼1 nm), more hydrophilic,43 and adopts adefined helical structure that could give rise to specific bindinginteractions with cations. Site-specific cation binding has notbeen observed with the cations studied here,1 and the diversityof polyelectrolytes studied also argues against this hypothesis.Backbone hydrophobicity is correlated with precipitateformation for synthetic sulfonate polyanions,44 but the oppositetrend is observed here. The methylphosphonate (MP)substitutions were designed to decouple DNA persistencelength and charge density. We are not aware of any directmeasurements of persistence length for MP nucleic acids, butsubstantial evidence indicates that DNA flexibility is determinedprimarily by base stacking.45,46 This is largely unperturbed byMP substitution,47 implying minimal difference in flexibilitybetween the MP and DNA helices. An upper bound can beestimated from EPR measurements,48 which show an increaseof as much as 40% in P−O bond mobility for MP vs DNA.Assuming a similar decrease in bending rigidity (i.e., no

Figure 4. Effects of nucleic acid backbone chemistry on complexphase. (A) Sequence of 21 nt luciferase siRNA from Hayashi et al.27

(B) Methylphosphonate substitution (right) at 10 of 22 nt reduces thecharge of the oligonucleotides by 45% compared to native DNA (samesequence as Figures 1−3). (C) When mixed with 50 aa pLys, ssRNA(left) forms coacervates while dsRNA (right) forms solid precipitates,just as for DNA. (D) When mixed with 50 aa pLys, mDNA−DNA*duplexes form precipitates like native DNA, while mDNA−mDNA*duplexes form coacervate complexes.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1636

stiffening from base stacking) and neglecting that 55% of theMP helix remains native DNA, we can estimate a persistencelength of at least ∼30 nm, which is still several times larger thanthe 7.5 nm axial length of the duplex−it should behave as arigid rod. This suggests charge density as the primarydeterminant of complex phase, a result consistent with ourand others’ observations49 of a precipitate−coacervate tran-sition as increasing counterion concentration decreases thestrength of electrostatic interactions.Our results provide insight into two aspects of polyelec-

trolyte complexation that are not well understood: theprecipitate/coacervate division, and mixtures with unevenpolymer lengths and charge stoichiometry. Few theoreticalapproaches address the former: mean-field theories of theVoorn-Overbeek type and random phase approximation-basedmodels predict conditions for phase separation but do notdifferentiate between coacervates and solid complexes.31,32,50

Schlenoff and co-workers have developed an appealing modelbased on the fraction of polyion charges paired with low-valence counterions that is qualitatively consistent with theobserved solid−liquid transition with increasing salt, as well asthe observed swelling of the coacervates, but requires empiricalparameters for each polyelectrolyte/counterion combination.49

Most models also assume either equal length polyelectrolytesor that one is much larger than the other, as well as bulkneutrality. Our measurements show that complexes formreadily over nearly 2 orders of magnitude in both polymerlength and charge ratios, and the uncomplexed DNAmeasurements (Figure 1B) indicate that the complexes arevery nearly neutral, in apparent conflict with models thatpredict overcharging by the more-abundant polyelectrolyte.51,52

The observation of qualitatively identical phase behavior fromthe shortest (N = 10) to longest polymers (N = 100) studiedhere, as well as the polyamines, is also interesting, as severalexperimental and theoretical studies suggest a qualitativedifference between short and long polyions, with only thelatter capable of true polyelectrolyte behavior.22,51,53 We arepresently undertaking measurements with even shorterpeptides to probe the breakdown of polyelectrolyte behaviorfurther.As discussed above, polyelectrolyte complexes are attractive

delivery vehicles for therapeutic RNA oligonucleotides, but theliterature is conflicted on the complexation properties of RNA,particularly duplexes such as siRNA. In particular, two reportssuggested that double-stranded RNA is incapable of formingpolyelectrolyte complexes with pLys and other cationicpeptides due to its larger bending rigidity compared to single-stranded RNA,15,16 while a third reported effective complex-ation.17 As shown in Figure 4, the Fluc dsRNA sequence usedby Hayashi et al. readily forms complexes with 50 aa pLys inour hands, though the complexes are solids rather than liquidsas for ssRNA. Their concentrations (up to 10 mg/mL, or ∼30mM charge)15 are comparable to ours, which suggests thateither solid-phase complexes are incompatible with micelleformation or that the PEG block inhibits complex formation.We have observed that polyelectrolyte complex micelles withsolid cores can be formed from homochiral peptides, thoughthe rate is much slower than for micelles with liquid cores.33

We are presently conducting experiments to test whether thePEG block inhibits complex formation in some way specific tonucleic acids; resolving this question is crucial for applyingpolyelectrolyte complexes to nucleic acid delivery. The abilityto explore a diverse range of complex properties also suggests

that oligonucleotide/oligopeptide systems can provide impor-tant insights into intracellular phase separation by exploring theroles of RNA structure and peptide sequence.

■ MATERIALS AND METHODSThe DNA sequence complementary to human microRNA-21(TCAACATCAGTCTGATAAGCTA) was used as a basis forconstructing a family of oligonucleotides of various lengths, withextensions and other oligonucleotides designed using the NUPACKsoftware tool54 to avoid unwanted secondary structures and self-dimerization. DNA and RNA oligonucleotides were purchased fromIntegrated DNA Technologies and the methylphosphonate oligonu-cleotides were purchased from TriLink BioTechnologies. Sequencesare listed in SI Tables S1−S6. Poly(L)lysine peptides (pLys, 10−100aa) were purchased from Alamanda Polymers as the chloride salt(bromide salt for P(D)K100). Prior to use, the peptides wereneutralized with NaOH and resuspended in water at 10 mM charge(NH2) concentration. Synthesis and characterization of the (KG)15and (KGG)10 peptides are described in SI Section S1, along withcomplete experimental details.

Polyelectrolyte complexes were prepared at pH 7 and roomtemperature (RT). Double-stranded DNA was prepared by annealingcomplementary strands at 45 °C for 5 min followed by slow cooling toRT. 18.2 MΩ water and concentrated NaCl solutions were mixed,followed by addition of the nucleic acid and then the peptide orpolyamine. Samples were mixed thoroughly after addition of eachpolyelectrolyte. For the charge ratio measurements (Figure 1), totalcharge concentration ([DNA phosphate] + [pLys amine]) was fixed at5 mM. In all other experiments, the charge concentration of eachpolyelectrolyte was 2.5 mM unless otherwise specified.

Phase and morphology of the complexes were observed by bright-field and phase contrast optical microscopy using a Leica DMI-6000Binverted microscope with white light illumination and 5−20×magnification. 100 μL aliquots of the complex suspensions wereplaced in ultralow attachment 96 well plates (Costar, Corning). Imageswere taken shortly after mixing and then again 4 h later, with the latterused unless noted to the contrary. Coacervate droplet sizes weredetermined using ImageJ 64; detailed analysis methodology can befound in SI Section S1.3. Fluorescence microscopy data were acquiredusing an Olympus DSU spinning disk microscope in wide field mode.Fluorescein- and TAMRA-labeled oligonucleotides (SI Section S1.6)were doped into unlabeled oligos at 10% (fluorescein) and 1%(TAMRA) ratios; complexes were otherwise prepared identically tothe other experiments.

To determine the amount of DNA remaining in solution aftercomplex formation, 100 μL aliquots were centrifuged for 10 min at20 000g to collect complexes at the bottom of 1.5 mL microcentrifugetubes. 50 μL of supernatant was removed from the top of the solutionand diluted 4× with 5 M NaCl solution. DNA was quantified byabsorbance at 260 nm using nearest-neighbor extinction coefficientsprovided by IDT. Typical fractions of DNA recovered under high-saltconditions where complexes were not seen were 80−90%, with noapparent dependence on the length of either polymer. For the chargeratio results in Figure 1B, the fraction of soluble DNA was normalizedby comparison to the DNA-only (defined as 100% free) and peptide-only (ADNA = 0) mixtures. The data in Figure 2D was normalized tothe average fraction of DNA recovered from these samples at 1 MNaCl (78.7% of the nominal input; the 50 aa complexes may not betotally dissociated) to aid visual comparison.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/jacs.7b03567.

Additional meterials, methods, and results (Figures S1−S21 and Tables S1−S7), and additional references(PDF)Coacervate coalescence.mov (AVI)

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1637

Precipitate Coalescence.mov (AVI)NaCl melting movie (AVI)Non-complementary fluor movie (AVI)Complementary fluor movie (AVI)

■ AUTHOR INFORMATIONCorresponding Author*mtirrell@uchicago.eduORCIDJeffrey R. Vieregg: 0000-0002-4786-5867Michael Lueckheide: 0000-0001-8310-0557NotesThe authors declare no competing financial interest.#These authors conttributed equally to this work.

■ ACKNOWLEDGMENTSThis work was supported by the U.S. Department of EnergyOffice of Science, Program in Basic Energy Sciences, MaterialsSciences and Engineering Division.

■ REFERENCES(1) Bloomfield, V. A. Curr. Opin. Struct. Biol. 1996, 6 (3), 334−341.(2) Teif, V. B.; Bohinc, K. Prog. Biophys. Mol. Biol. 2011, 105 (3),208−222.(3) Shin, Y.; Brangwynne, C. P. Science 2017, 357, 6357.(4) Banani, S. F.; Lee, H. O.; Hyman, A. A.; Rosen, M. K. Nat. Rev.Mol. Cell Biol. 2017, 18 (5), 285−298.(5) Hnisz, D.; Shrinivas, K.; Young, R. A.; Chakraborty, A. K.; Sharp,P. A. Cell 2017, 169 (1), 13−23.(6) Jain, A.; Vale, R. D. Nature 2017, 546 (7657), 243−247.(7) Vieregg, J. R.; Tang, T. Y. D. Curr. Opin. Colloid Interface Sci.2016, 26, 50−57.(8) Lachelt, U.; Wagner, E. Chem. Rev. 2015, 115 (19), 11043−11078.(9) Ponnuswamy, N.; Bastings, M. M. C.; Nathwani, B.; Ryu, J. H.;Chou, L. Y. T.; Vinther, M.; Li, W. A.; Anastassacos, F. M.; Mooney,D. J.; Shih, W. M. Nat. Commun. 2017, 8, 1−9.(10) Tian, T.; Zhang, T.; Zhou, T.; Lin, S.; Shi, S.; Lin, Y. Nanoscale2017, 485, 623−11.(11) Voets, I. K.; de Keizer, A.; Cohen Stuart, M. A. Adv. ColloidInterface Sci. 2009, 147−148, 300−318.(12) Miyata, K.; Nishiyama, N.; Kataoka, K. Chem. Soc. Rev. 2012, 41(7), 2562.(13) Kuo, C.-H.; Leon, L.; Chung, E. J.; Huang, R.-T.; Sontag, T. J.;Reardon, C. A.; Getz, G. S.; Tirrell, M.; Fang, Y. J. Mater. Chem. B2014, 2, 8142−8153.(14) Scholz, C.; Wagner, E. J. Controlled Release 2012, 161 (2), 554−565.(15) Hayashi, K.; Chaya, H.; Fukushima, S.; Watanabe, S.; Takemoto,H.; Osada, K.; Nishiyama, N.; Miyata, K.; Kataoka, K. Macromol. RapidCommun. 2016, 37 (6), 486−493.(16) Kwok, A.; McCarthy, D.; Hart, S. L.; Tagalakis, A. D. Chem. Biol.Drug Des. 2016, 87 (5), 747−763.(17) Chou, S.-T.; Hom, K.; Zhang, D.; Leng, Q.; Tricoli, L. J.;Hustedt, J. M.; Lee, A.; Shapiro, M. J.; Seog, J.; Kahn, J. D.; Mixson, A.J. Biomaterials 2014, 35 (2), 846−855.(18) Record, M. T.; Anderson, C. F.; Lohman, T. M. Q. Rev. Biophys.1978, 11 (2), 103−178.(19) Mascotti, D. P.; Lohman, T. M. Proc. Natl. Acad. Sci. U. S. A.1990, 87 (8), 3142−3146.(20) Laemmli, U. K. Proc. Natl. Acad. Sci. U. S. A. 1975, 72 (11),4288−4292.(21) Maruyama, A.; Katoh, M.; Ishihara, T.; Akaike, T. BioconjugateChem. 1997, 8 (1), 3−6.(22) Ballin, J. D.; Shkel, I. A.; Record, M. T. Nucleic Acids Res. 2004,32 (11), 3271−3281.

(23) Shkel, I. A.; Record, M. T. Soft Matter 2012, 8 (36), 9345−11.(24) Li, L.; Pabit, S. A.; Meisburger, S. P.; Pollack, L. Phys. Rev. Lett.2011, 106 (10), 108101−108104.(25) Katz, A. M.; Tolokh, I. S.; Pabit, S. A.; Baker, N.; Onufriev, A.V.; Pollack, L. Biophys. J. 2017, 112 (1), 22−30.(26) de Jong, H. B.; Kruyt, H. R. Proc. Koninklijke NederlandseAkademie Wetenschappen 1929, 32, 849−856.(27) de Kruif, C. G.; Weinbreck, F.; de Vries, R. Curr. Opin. ColloidInterface Sci. 2004, 9 (5), 340−349.(28) Priftis, D.; Laugel, N.; Tirrell, M. Langmuir 2012, 28 (45),15947−15957.(29) Fu, J.; Schlenoff, J. B. J. Am. Chem. Soc. 2016, 138 (3), 980−990.(30) Gouin, S. Trends Food Sci. Technol. 2004, 15 (7−8), 330−347.(31) Kudlay, A.; Ermoshkin, A. V.; Olvera de la Cruz, M.Macromolecules 2004, 37 (24), 9231−9241.(32) Spruijt, E.; Westphal, A. H.; Borst, J. W.; Cohen Stuart, M. A.;van der Gucht, J. Macromolecules 2010, 43 (15), 6476−6484.(33) Perry, S. L.; Leon, L.; Hoffmann, K. Q.; Kade, M. J.; Priftis, D.;Black, K. A.; Wong, D.; Klein, R. A.; Pierce, C. F.; Margossian, K. O.;Whitmer, J. K.; Qin, J.; de Pablo, J. J.; Tirrell, M. Nat. Commun. 2015,6, 1−8.(34) Santalucia, J.; Hicks, D. Annu. Rev. Biophys. Biomol. Struct. 2004,33 (1), 415−440.(35) Vieregg, J.; Cheng, W.; Bustamante, C.; Tinoco, I. J. Am. Chem.Soc. 2007, 129 (48), 14966−14973.(36) Chollakup, R.; Smitthipong, W.; Eisenbach, C. D.; Tirrell, M.Macromolecules 2010, 43 (5), 2518−2528.(37) Priftis, D.; Tirrell, M. Soft Matter 2012, 8 (36), 9396−9405.(38) Shapiro, J. T.; Leng, M.; Felsenfeld, G. Biochemistry 1969, 8 (8),3219−3232.(39) Porschke, D. Biophys. Chem. 1979, 10 (1), 1−16.(40) Aumiller, W. M.; Keating, C. D. Nat. Chem. 2016, 8 (2), 129−137.(41) Aumiller, W. M.; Pir Cakmak, F.; Davis, B. W.; Keating, C. D.Langmuir 2016, 32 (39), 10042−10053.(42) Yin, Y.; Niu, L.; Zhu, X.; Zhao, M.; Zhang, Z.; Mann, S.; Liang,D. Nat. Commun. 2016, 7, 1−7.(43) Costa, D.; Miguel, M. G.; Lindman, B. J. Phys. Chem. B 2007,111 (37), 10886−10896.(44) Fu, J.; Fares, H. M.; Schlenoff, J. B.Macromolecules 2017, 50 (3),1066−1074.(45) Mills, J. B. Nucleic Acids Res. 2004, 32 (13), 4055−4059.(46) McIntosh, D. B.; Duggan, G.; Gouil, Q.; Saleh, O. A. Biophys. J.2014, 106 (3), 659−666.(47) Kan, L. S.; Cheng, D. M.; Miller, P. S.; Yano, J.; Ts'o, P. O. P.Biochemistry 1980, 19 (10), 2122−2132.(48) Okonogi, T. M.; Alley, S. C.; Harwood, E. A.; Hopkins, P. B.;Robinson, B. H. Proc. Natl. Acad. Sci. U. S. A. 2002, 99 (7), 4156−4160.(49) Wang, Q.; Schlenoff, J. B. Macromolecules 2014, 47 (9), 3108−3116.(50) Raspaud, E.; Olvera de la Cruz, M.; Sikorav, J. L.; Livolant, F.Biophys. J. 1998, 74 (1), 381−393.(51) Zhang, R.; Shklovskii, B. I. Phys. A 2005, 352 (1), 216−238.(52) Oskolkov, N. N.; Potemkin, I. I. Macromolecules 2007, 40 (23),8423−8429.(53) Toma, A. C.; de Frutos, M.; Livolant, F.; Raspaud, E.Biomacromolecules 2009, 10 (8), 2129−2134.(54) Zadeh, J. N.; Steenberg, C. D.; Bois, J. S.; Wolfe, B. R.; Pierce,M. B.; Khan, A. R.; Dirks, R. M.; Pierce, N. A. J. Comput. Chem. 2011,32 (1), 170−173.

Journal of the American Chemical Society Article

DOI: 10.1021/jacs.7b03567J. Am. Chem. Soc. 2018, 140, 1632−1638

1638